Astrocyte elevated gene-1 (AEG-1) functionsas an oncogene and regulates angiogenesisLuni Emdada,1, Seok-Geun Leeb,c,1, Zhao Zhong Sub, Hyun Yong Jeonb, Habib Boukerched, Devanand Sarkarb,c,e,and Paul B. Fisherb,c,e,2

aDepartment of Neurosurgery and Oncological Sciences, Mount Sinai School of Medicine, New York, NY 10029; bDepartment of Human and MolecularGenetics, cVirginia Commonwealth University Massey Cancer Center, eVirginia Commonwealth University Institute of Molecular Medicine, VirginiaCommonwealth University, School of Medicine, Richmond, VA 23298; and dEA 4174, University Lyon 1, French National Institute for Health andMedical Research, France

Communicated by George J. Todaro, Targeted Growth, Inc., Seattle, WA, October 7, 2009 (received for review March 17, 2009)

Astrocyte-elevated gene-1 (AEG-1) expression is increased in mul-tiple cancers and plays a central role in Ha-ras-mediated oncogen-esis through the phosphatidylinositol 3-kinase (PI3K)/Akt signalingpathway. Additionally, overexpression of AEG-1 protects primaryand transformed human and rat cells from serum starvation-induced apoptosis through activation of PI3K/Akt signaling. Thesefindings suggest, but do not prove, that AEG-1 may function as anoncogene. We now provide definitive evidence that AEG-1 isindeed a transforming oncogene and show that stable expressionof AEG-1 in normal immortal cloned rat embryo fibroblast (CREF)cells induces morphological transformation and enhances invasionand anchorage-independent growth in soft agar, two fundamentalbiological events associated with cellular transformation. Addi-tionally, AEG-1-expressing CREF clones form aggressive tumors innude mice. Immunohistochemistry analysis of tumor sections dem-onstrates that AEG-1-expressing tumors have increased microves-sel density throughout the entire tumor sections. Overexpressionof AEG-1 increases expression of molecular markers of angiogen-esis, including angiopoietin-1, matrix metalloprotease-2, and hy-poxia-inducible factor 1-�. In vitro angiogenesis studies furtherdemonstrate that AEG-1 promotes tube formation in Matrigel andincreases invasion of human umbilical vein endothelial cells via thePI3K/Akt signaling pathway. Tube formation induced by AEG-1correlates with increased expression of angiogenesis markers,including Tie2 and hypoxia-inducible factor-�, and blocking AEG-1-induced Tie2 with Tie2 siRNA significantly inhibits AEG-1-inducedtube formation in Matrigel. Overall, our findings demonstrate thataberrant AEG-1 expression plays a dominant positive role inregulating oncogenic transformation and angiogenesis. Thesefindings suggest that AEG-1 may provide a viable target for directlysuppressing the cancer phenotype.

angiogenesis-related molecules � tumor progression

Cancer development and acquisition of malignant potentialare multifactor and multistep processes that occur in a

temporal manner during tumor progression (1, 2). Significantcomponents of these processes in the evolving neoplastic cellinclude development of growth signal autonomy, insensitivity togrowth-inhibitory signals, evasion from apoptosis, unlimitedreplicative potential, and aberrant angiogenesis, as well as tissueinvasion that can ultimately culminate in tumor dissemination(1–3). An important defining element in controlling growth inboth primary and metastatic tumors is new blood vessel forma-tion (angiogenesis). The importance of angiogenesis in thegrowth of solid tumors is well established (3, 4). Indeed, tumorsize is restricted to a few cubic millimeters if it is not able toattract new blood vessels. Increased intratumoral microvasculardensity relative to normal tissue is observed in tumors ofdifferent tissues including the brain, colon, and breast (5–7).Production of new vessels by the developing tumor and distantmetastases results from the amplification of large quantities ofpro-angiogenic molecules by both the tumor and host cells and

reflects a net balance between positive and negative regulatorsof angiogenesis (8–10). These observations emphasize that anygenetic modification(s) in a cancer cell that culminates inexpansion of tumor growth and metastasis will be inexorablylinked to angiogenesis.

Astrocyte elevated gene (AEG)-1 was cloned by rapid subtrac-tion hybridization as a gene induced in primary human fetalastrocytes (PHFA) infected with HIV-1 or treated with tumornecrosis factor-� (TNF-�) (11). Intriguingly, expression analysisrevealed that AEG-1 was significantly elevated in subsets of breastcarcinoma, melanoma, and malignant glioma cell lines compared totheir normal cellular counterparts (12). Its expression was alsoelevated in adult astrocytes transformed by SV40-T antigen, humantelomerase reverse transcriptase (hTERT), and oncogenic Ha-rasthat displayed an aggressive glioma-like phenotype (12). AEG-1synergized with oncogenic Ha-ras to enhance soft agar colonyformation of SV40-T antigen-induced immortalized human mela-nocytes (FM516-SV) as well as in PHFA. Interestingly, AEG-1itself is a downstream target of Ha-ras and plays an important rolein mediating the growth-promoting effects of Ha-ras (13). Over-expression of AEG-1 also augmented the anchorage-independentgrowth of HeLa cells and human glioma cell lines and increasedtheir migration and invasion properties (14, 15). Conversely, inhi-bition of AEG-1 by siRNA significantly inhibited migration andinvasion of malignant glioma cells and prostate cancer cells and invivo lung metastasis of breast cancer cells (16–19). In PHFA,FM516-SV, and cloned rat embryo fibroblasts (CREF), AEG-1protects from serum starvation-induced apoptosis by activating thePI3K/Akt signaling pathway indicating, but not definitively proving,that AEG-1 might function as an oncogene (20). Inhibition ofAEG-1 in prostate cancer cells downregulated Akt activation andlead to upregulation of forkhead box (FOXO) 3a activity resultingin apoptosis (17). Other studies identified AEG-1 homologs in ratand mouse, named Lyric/3D3 and Metadherin, respectively (18, 21,22), and provided evidence for its involvement in cancer progres-sion and metastasis. In total, these observations indicate thatAEG-1 may represent an essential gene regulating multiple signal-ing and biochemical pathways leading to cell transformation andtumor progression in diverse target cells.

A number of questions remain concerning the potentialrole(s) of AEG-1 in regulating the cancerous state and preciselywhat phenotypes it impacts on. These include foremost thebiological consequences of elevated AEG-1 expression in nor-mal cells in vitro and in vivo and its potential definitive role, ifany, in angiogenesis, which as indicated above is a hallmark of

Author contributions: L.E., S.-G.L., D.S., and P.B.F. designed research; L.E., S.-G.L., Z.Z.S., andH.Y.J. performed research; L.E., S.-G.L., Z.Z.S., H.B., D.S., and P.B.F. contributed new re-agents/analytic tools; L.E., S.-G.L., Z.Z.S., H.Y.J., H.B., D.S., and P.B.F. analyzed data; and L.E.,S.-G.L., and P.B.F. wrote the paper.

The authors declare no conflict of interest.

1L.E. and S.-G.L. contributed equally to this work.

2To whom correspondence should be addressed. E-mail: pbfisher@vcu.edu.

21300–21305 � PNAS � December 15, 2009 � vol. 106 � no. 50 www.pnas.org�cgi�doi�10.1073�pnas.0910936106

Dow

nloa

ded

by g

uest

on

Janu

ary

27, 2

021

cancer progression and metastasis. In this study, we demonstratethat AEG-1 can function as an oncogene, which when expressedat physiological levels in normal immortal CREF cells results inmorphological transformation, enhanced invasion, anchorage-independent growth in agar, and acquisition of tumorigenicpotential when injected into athymic nude mice. Additionally,sections of AEG-1-overexpressing tumors showed enhancedCD31 expression indicating that CREF-AEG-1 tumors arehighly vascularized. In vitro angiogenesis assays with humanumbilical vein endothelial cells (HUVECs) revealed that over-expression of AEG-1 significantly increased tube formation inMatrigel through PI3K/Akt signaling. Furthermore, overexpres-sion of AEG-1 in HUVECs and malignant glioma cells modu-lated angiogenic regulators including Tie2 and hypoxia induciblefactor 1 (HIF1)-�. In total, the present studies now confirm thatAEG-1 is indeed an oncogene and also a direct regulator ofangiogenesis by upregulating key components in the process ofblood vessel formation.

ResultsAEG-1 Induces Invasion and Anchorage-Independent Growth in Nor-mal Immortal CREF Cells. Anchorage-independent growth andinvasion are two crucial events in tumor initiation and progres-sion. Aberrant expression of AEG-1 modulates these phenotypesin HeLa, malignant glioma, prostate cancer, neuroblastoma, andhepatocellular carcinoma cells (14, 15, 17, 23, 24). AEG-1 alsosynergizes with Ha-ras to augment the transformed phenotype inFM-516-SV and PHFA cells (12). However, the oncogenicpotential of AEG-1 as a single gene in nontransformed cellsremained to be determined. To explore this possibility, CREFcells were engineered to stably express AEG-1 (Fig. 1). We usedCREF cells for this study because they can be morphologicallytransformed by single oncogenes, including Ha-ras, Ad E1A,v-Src, human papilloma virus type 18, and v-Raf, or highmolecular weight human tumor-derived DNA (25, 26), and can

therefore serve as a sensitive barometer for monitoring onco-gene functions (25–27). As shown in Fig. 1, parental CREF donot form macroscopic colonies in soft agar (Fig. 1B) and areminimally invasive (Fig. 1 C and D), whereas AEG-1-transformed clones all grow with variable efficiencies in soft agarand are highly invasive. Expression levels of AEG-1 positivelycorrelated with agar cloning efficiency and invasive abilities ofCREF-AEG-1 stable clones, suggesting a central role of AEG-1in development of the transformed state. A concern whenexamining the function of transfected genes in cells is their levelof expression and how this relates to true endogenous geneexpression in untransfected cells. A comparison of the level ofAEG-1 protein expressed endogenously in a series of humancancers and normal human cells indicated that the level ofexpression of AEG-1 in stable CREF-AEG-1 clone 2 and clone30 is comparable or lower than observed in normal immortalhuman astrocyte, prostate, and melanocyte cells (Fig. 1 A). Incontrast, AEG-1 expression is significantly lower in these trans-formed CREF clones than in human cancers, including malig-nant gliomas, prostate cancers, and melanomas. These findingsfurther support genuine oncogenic functions attributed to ex-pression of human AEG-1 at physiologically relevant levels inCREF cells.

AEG-1 as a Single Gene Promotes Tumor Formation When Expressedin CREF Cells. To evaluate oncogenic potential of cells expressingAEG-1 in vivo, nude mice were s.c. injected with CREF-AEG-1clones (Fig. 2). Inoculation of nude mice with CREF-AEG-1cells (five independent clones of CREF genetically engineered toexpress elevated levels of AEG-1) resulted in the formation ofaggressive and highly vascularized tumors (Figs. 2 and 3A). Incontrast, as previously observed (26, 28), injection of nude micewith parental CREF cells did not generate any visible tumors.These in vivo results document that AEG-1 can function as anoncogene when expressed as a single genetic element at physi-ologically pertinent levels in immortal normal rat embryo cellsresulting in an aggressive tumorigenic phenotype in nude mice.

AEG-1 Expression Directly Correlates with Increased Expression ofAngiogenesis Markers and Elevated Angiogenesis. On the basis ofthe morphology (high vascularization) of tumors in nude miceinduced by CREF-AEG-1 cells (Fig. 3A), we focused on angio-

Fig. 1. Effects of stable overexpression of AEG-1 in CREF cells on colonyformation in soft agar and cell invasion. (A) CREF cells were stably transfectedwith either the empty pcDNA3.1 vector or the AEG-1 expression vector,respectively. CREF-AEG-1 clones were selected for expression of AEG-1. (Left )Expression of AEG-1 protein by stably transfected CREF-AEG-1 cells is shown byWestern blot analysis. EF1� was used as an internal control to ascertain equalloading. (Right) Expression of AEG-1 protein in stably transfected CREF-AEG-1cells (clone 2 and 30) and normal immortal CREF and a series of normal humancells (PHFA, P69, FM516-SV) and corresponding human tumor cells (U87MG,H4, DU-145, PC-3, HO-1, C8161) by Western blot analysis. EF1� was used as aninternal control for protein loading. (B) A total of 1 � 105 cells were seeded in0.4% agar on 0.8% base agar. Two weeks later, colonies �0.1-mm werecounted under a dissection microscope. *, P � 0.05 vs. CREF. (C) Cells (5 � 104)were seeded onto the upper chamber of a Matrigel invasion chamber systemin the absence of serum. Twenty-four hours after seeding, the filters werefixed, stained, and photographed. (D) Quantitation of the invasion assay. Thedata expressed in the graph is the mean � SE of three independent experi-ments. *, P � 0.05 vs. CREF.

Fig. 2. Overexpression of AEG-1 in CREF cells induces tumorigenesis in vivo.(A) 2 � 106 cells of each cell line were subcutaneously injected into the flankof each nude mouse. Tumor volumes were measured at the indicated timepoints. Results are expressed as means � SE (n � 5 per group). CREF did notform tumors in nude mice. The average tumor volume in cubic millimeters offive animals � SD. Unpaired two-tailed Student’s t test (P � 0.01). (B) Photo-graph showing tumor growth in nude mice. (C) Tumor weights were measuredafter sacrifice of the mice 4 weeks after injection. The data represent mean �SE in each group (*, P � 0.01 vs. CREF).

Emdad et al. PNAS � December 15, 2009 � vol. 106 � no. 50 � 21301

MED

ICA

LSC

IEN

CES

Dow

nloa

ded

by g

uest

on

Janu

ary

27, 2

021

genesis as a possible process underlying the increased in vivoaggressiveness of these transformed cells resulting from en-hanced AEG-1 expression. Tumors were isolated, formalin-fixed, and paraffin-embedded sections were made. Staining forCD31 demonstrated the presence of microvessels indicative ofincreased angiogenesis (Fig. 3B). AEG-1 staining was mostprominent in the peri-nuclear region of the cells, with only aminority of tumor cells also showing nuclear staining (Fig. 3C).Additional immunohistochemical analysis revealed that en-hanced expression of AEG-1 in tumor sections augmentedexpression of specific angiogenesis molecules including angio-poietin-1 (Ang1), matrix metalloprotease (MMP)-2, and HIF1-�(Fig. 3C), further supporting a potential role of AEG-1 in tumorangiogenesis.

AEG-1 Functions as a Potent Inducer of Angiogenesis. Histochemicalanalysis of the CREF-AEG-1 tumors indicated potential angio-genic activities of AEG-1, which we investigated using in vitroangiogenesis assays. Tube formation is an important parameterof endothelial function in angiogenesis that can readily beevaluated and quantified in vitro. HUVECs cultured on Matrigelrapidly align, extend processes into the matrix, and finally formcapillary-like structures surrounding a central lumen. Within16 h after seeding on Matrigel, HUVECs start to form tubularstructures (Fig. 4A). Forced expression of AEG-1 by a replica-tion incompetent adenovirus (Ad.AEG-1) significantly en-hanced tube formation, with an increased number of intracel-lular contacts and overall complexity of the network (Fig. 4A).Conversely, treatment of HUVECs with AEG-1 siRNA signif-icantly inhibited VEGF-induced tube formation in Matrigel (Fig.4B). Next we used a chicken chorioallantoic membrane (CAM)angiogenesis model to study whether siRNA knockdown ofAEG-1 would be a feasible approach to suppress tumor-inducedangiogenesis in vivo. H4 glioma cells, previously shown toexpress elevated levels of AEG-1, were treated with eithercontrol or AEG-1 siRNA and seeded on fresh, 9-day-old CAMs.One week after seeding the cells, the CAMs were harvested.CAMs seeded with control siRNA-treated H4 cells causedoutgrowth capillary neovascularization near the tumor coreregion (Fig. 4C, left). On the other hand, CAMs seeded withAEG-1 si-RNA-treated H4 glioma cells exhibited little neovas-cularization (Fig. 4C, right).

Previous studies have linked many AEG-1-induced cellularchanges with PI3K/Akt signaling (13, 20). PI3K/Akt regulates anumber of processes associated with cancer progression includ-ing angiogenesis (29). To analyze the potential importance ofPI3K/Akt signaling in AEG-1-mediated angiogenesis, we used arecombinant replication-incompetent adenovirus expressing adominant-negative Akt (Ad.DN.Akt). AEG-1-mediated endo-thelial cell tube formation was significantly inhibited byAd.DN.Akt (Fig. 4A), suggesting that the PI3K/Akt pathway isan integral component of this process.

Since endothelial cell migration and invasion are importantsteps in angiogenesis, we tested the effect of forced expressionof AEG-1 on the invasive ability of HUVECs. As shown in Fig.4D, overexpression of AEG-1 significantly enhanced the invasiveability of HUVECs, which was conversely blocked by inhibitionof the PI3K/Akt pathway. Taken together, these results suggestthat AEG-1 is a potent inducer of angiogenesis, and PI3K/Aktsignaling is a necessary event in AEG-1-induced angiogenesis.

AEG-1 Modulates Angiogenic Regulators. To further explore themechanism(s) that mediate the pro-angiogenic activities of

Fig. 3. Histochemical analysis of tumors derived from nude mice injectedwith AEG-1 stable overexpressing CREF clone 2 and 30. (A) Representativeexamples of tumors derived from CREF-AEG-1 clone 2 and 30. (B) The sectionsof CREF-AEG-1 clone 2 and 30 xenografts and human glioma tissue wereimmunostained for the endothelial cell marker CD31. Immunohistochemicalanalysis demonstrated robustly increased blood vessel density in CREF-AEG-1tumor sections (inserts, high magnification view). (C) Serial sections of forma-lin-fixed, paraffin-embedded tumors were stained with a rabbit polyclonalAEG-1, a rabbit polyclonal MMP-2, a rabbit polyclonal Ang1, and a mousemonoclonal HIF1-� antibodies. Signals were developed with DAB chromogen(brown, AEG-1 and MMP-2) or Vector VIP (purple, Ang1 and HIF1-�) andcounterstained with hematoxylin.

Fig. 4. AEG-1 promotes angiogenesis. (A) AEG-1 induces tube formation onMatrigel through the PI3K/Akt signaling pathway. HUVECs were infected withAd.vec or Ad.AEG-1 (25 PFU/cell) in combination with Ad.DN.Akt (25 PFU/cell).One day after infection, cells (5 �104), which were labeled with a fluorescentdye, calcein AM, were seeded onto Matrigel and tube formation was assayedafter 16 h by fluorescence microscopy. Upper, cells that were infected withAd.AEG-1 clearly showed an increase in tube formation on Matrigel (arrows),whereas the Ad.vec-treated cells were a poor inducer of tube-like structures(arrowheads). (Lower) Graphical presentation of tube formation assay, dataexpressed in the graph is the mean � SE of three independent experiments. *,P � 0.05 vs. Ad.vec-infected cells; #, P � 0.05 vs. Ad.AEG-1-infected cells. (B)HUVECs were treated with either control siRNA or AEG-1 siRNA, plated onMatrigel and stimulated with VEGF (10 ng/mL). Tube formation was assayedafter 16 h by fluorescence microscopy. *, P � 0.05 vs. control siRNA treatedcells. (C) Inhibition of AEG-1 by siRNA inhibits angiogenesis in in vivo CAMassays. CAMs of 9-day-old chicken embryos were injected with either controlsiRNA or AEG-1 siRNA-transfected H4 glioma cells. Angiogenesis in yolk sacwas monitored 1 week after inoculation, and representative fields werephotographed. (D) HUVECs were infected with Ad.vec or Ad.AEG-1 (25 PFU/cell) in combination with Ad. DN.Akt (25 PFU/cell). One day after infection,cells (5 � 104) were seeded onto the upper chamber of a Matrigel invasionchamber system in the absence of serum. Twenty-four hours after seeding, thefilters were fixed, stained, and photographed. (Right panel) Graphical repre-sentation of the invasion assay. The data expressed in the graph is the mean �SE of three independent experiments. *, P � 0.05 vs. Ad.vec-infected cells, #P �0.05 vs. Ad.AEG-1-infected cells.

21302 � www.pnas.org�cgi�doi�10.1073�pnas.0910936106 Emdad et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

27, 2

021

AEG-1, we investigated the effects of AEG-1 on Tie2 andHIF1-� expressions, which are commonly associated with an-giogenesis. As shown in Fig. 5A, overexpression of AEG-1significantly augmented the expressions of Tie2 and HIF1-� inHUVECs. Additionally, overexpression of AEG-1 significantlyaugmented the expression of HIF1-� in U87 and H4 humanmalignant glioma cells. This modulation of angiogenic regulatorsby AEG-1 was abrogated by co-expression of a dominant neg-ative Akt, further supporting the involvement of PI3K/Aktsignaling in AEG-1-induced angiogenesis. To test the specificrole of Tie2 in the angiogenic activities of AEG-1, HUVECswere transfected with either control or Tie2 siRNA and infectedwith Ad.AEG-1 or an Ad vector not containing the AEG-1 gene(Ad.vec) as a control. Treatment with Tie2 siRNA significantlyabrogated AEG-1-induced tube formation in Matrigel (Fig. 5B),indicating that Tie2 plays a significant role in AEG-1-inducedangiogenesis. Since vascular endothelial growth factor (VEGF)is a known inducer of angiogenesis and the Tie2 and VEGFreceptor (VEGFR) pathways seem to work in a complementaryand coordinated fashion during vascular development, we ex-amined the effects of AEG-1 overexpression on VEGF promoteractivity. As shown in Fig. 5 C and D, aberrant expression ofAEG-1 in H4 cells modestly enhanced VEGF promoter activity,which was ablated by co-expression of a dominant-negative Aktor Tie2 siRNA, indicating the role of the PI3K/Akt pathway andTie2 in the AEG-1-mediated VEGF induction.

DiscussionCancer is a continuous process culminating in transformed cellsthat develop both qualitatively and quantitatively distinct prop-

erties as a tumor grows and evolves (30, 31). An initial step inconversion of a cell from normal to transformed can frequentlyinvolve the activation of a positive acting oncogene that signif-icantly alters cellular phenotype (30, 31). Additionally, it is nowconsidered almost axiomatic that tumor progression is mani-fested by the ability of the cancer cell to grow in an unrestrainedmanner and establish secondary colonies, both of which requirea well-developed set of blood vessels, i.e., angiogenesis, to sustainthe growth of the primary and secondary metastatic tumors (32,33). In the last few years, multiple lines of evidence havesuggested that AEG-1, a unique and highly conserved protein,plays a significant role in the progression of multiple cancers andmetastasis. In the present study, we provide definitive evidencethat AEG-1 can function directly as an oncogene and is also apotent inducer of angiogenesis.

Previous studies have demonstrated that AEG-1 can protectPHFA, rat embryo fibroblasts, and SV40-immortalized humanmelanocytes from serum-starvation-induced apoptosis (20).AEG-1 alone modestly increases soft agar growth of PHFA, butcan synergize with Ha-ras to significantly augment anchorage-independent growth (12). In addition, we now demonstrate thatAEG-1 as a single genetic element can significantly augmentanchorage-independent growth in normal immortal CREF cells(Fig. 1B) and can induce aggressive tumor growth in nude mice(Fig. 2). Tumors developing from injection with CREF-AEG-1clones are highly vascularized and express elevated levels ofCD31 (Fig. 3B), a marker of angiogenesis. CREF-AEG-1 tumorsalso display high levels of additional angiogenic markers, includ-ing Ang1, MMP-2, and HIF1-� (Fig. 3C). Moreover, we haverecently shown that overexpression of AEG-1 in nontumorigenichuman hepatocellular carcinoma cells also results in highlyaggressive and vascularized tumors (24). These findings providesupport for the hypothesis that AEG-1 might be a positiveregulator of angiogenesis. This is indeed the case, since we nowshow that overexpression of AEG-1 in HUVECs significantlyenhances tube formation in Matrigel (Fig. 4A); conversely,knockdown of AEG-1 using siRNA significantly abrogatesVEGF-induced tube formation in Matrigel (Fig. 4B).

To elucidate the precise molecular mechanism underlyingAEG-1 function as an angiogenesis inducer, we focused onexpression of Tie2, HIF1-�, and Ang1 in AEG-1-overexpressingHUVECs, U87, and H4 malignant glioma cells. Ang1 and Ang2function as ligands for Tie2, a tyrosine kinase receptor predom-inantly expressed in vascular endothelial cells. Ang1 specificallyinduces tyrosine phosphorylation of Tie2, which results in theactivation of multiple activities related to angiogenesis such asendothelial cell migration, tube formation, sprouting, and sur-vival (34, 35). Upregulation of Ang1 in high-grade gliomas,non-small cell lung carcinomas, and ovarian, breast, and gastriccarcinomas strongly correlate with tumor malignancy (36–40).Furthermore, overexpression of Ang1 in HeLa, GS9L, and insome established glioma cell lines has been reported to increasetumor growth (41–43). Increased Tie2 levels were found in thevasculature of a number of human tumors, including breastcancer, non-small cell lung cancer, hepatocellular carcinoma,prostate cancer, hemangioma, and astrocytoma, and these levelswere found to correlate with increasing malignancy (37, 44–47).Recent studies indicate that Tie2 plays a significant role inangiogenesis occurring in cancer (44–47). Suppression of Tie2signaling (47–49) using specific inhibitors, such as soluble dom-inant-negative receptors, an antisense oligonucleotide, RNAaptamers, RNA interference, or a short synthetic peptide,promotes antitumor effects by restraining tumor angiogenesis. Inthese contexts, modulation of Tie2 and Ang1 by AEG-1 (Fig. 5A)provides a mechanistic model by which AEG-1 regulates tumorangiogenesis (Fig. 6). We also show that AEG-1 enhancesHIF-1� expression in HUVECs, U87, and H4 human malignantglioma cells (Fig. 5A). HIF-1 is expressed in hypoxic tumor cells

Fig. 5. AEG-1 enhances expression of angiogenesis-associated genes andpromotes VEGF promoter activity. (A) HUVECs, U87 and H4 glioma cells wereinfected with the indicated virus as in Fig. 4. Forty-eight hours after infection,total cellular extracts were prepared in RIPA buffer. Equal amounts of proteinswere separated on 8–12% SDS-PAGE, transferred onto nitrocellulose mem-brane and probed with a chicken polyclonal AEG-1, a rabbit polyclonal Tie2and Ang1, and a mouse monoclonal HIF1-� antibodies. EF1� was used as acontrol to confirm equal protein loading. (B) HUVECs were transfected witheither control siRNA or Tie2 siRNA and then infected with Ad.vec or Ad.AEG-1(25 PFU/cell). One day after infection, cells (5 � 104) were seeded ontoMatrigel, and tube formation was assayed after 16 h. The data expressed in thegraph is the mean � SE of three independent experiments.

*, P � 0.05 vs. Ad.vec-infected cells; #, P � 0.05 vs. Ad.AEG-1-infected cells. (C)H4 cells were infected with the indicated virus as in Fig. 4. The next day cellswere transfected with pGL3/VEGF promoter and pSV-�-gal vectors, and 48 hlater, luciferase and �-gal activity were determined as described in Materialsand Methods. The data expressed in the graph is the mean � SE of threeindependent experiments. *, P � 0.05 vs. Ad.vec-infected cells; #, P � 0.05 vs.Ad.AEG-1-infected cells. (D) One day after infection with the indicated virus,H4 cells were transfected with pGL3/VEGF promoter and pSV-�-gal and eithercontrol or Tie2 siRNA. The data in the graph is the mean � SE of threeindependent experiments. *, P � 0.05 vs. Ad.vec-infected cells; #, P � 0.05 vs.Ad.AEG-1-infected cells.

Emdad et al. PNAS � December 15, 2009 � vol. 106 � no. 50 � 21303

MED

ICA

LSC

IEN

CES

Dow

nloa

ded

by g

uest

on

Janu

ary

27, 2

021

and transactivates various hypoxia-responsive genes, which con-fer malignant properties to tumors including apoptosis-resistance and enhanced tumor growth, invasion, and metastasis(50). In addition, HIF-1 activates pro-angiogenic cytokines suchas VEGF and platelet-derived growth factor, which increase theproliferation and re-growth of tumor blood vessels (50, 51). Wenow show that AEG-1 also activates the VEGF promoter inmalignant glioma cells (Fig. 5C), which is in agreement withobservations from our group in stable hepatocellular carcinomacells overexpressing AEG-1 that display elevated levels ofVEGF, placental growth factor (PIGF) and fibroblast growthfactor � (FGF�) compared to parental hepatocellular carcinomacells (24).

Our results also reveal the central contribution of PI3K/Aktsignaling in AEG-1-induced angiogenesis. Ang1 regulates endo-thelial cell survival through the PI3K/Akt pathway (52). Over-expression of a dominant-negative Akt construct inhibits Ang1-mediated endothelial cell survival (53), supporting a critical roleof Akt in Tie2/Ang-mediated endothelial cell survival. Addi-tionally, both Ang1- and VEGF-induced endothelial cell survivalare partially regulated through the PI3K/Akt pathway. In thesecontexts, our results firmly establish PI3K/Akt signaling as oneof the primary mediators of AEG-1-induced angiogenesis in bothnormal HUVECs and malignant glioma cells. The Ras/MEKsignaling pathway is another major pathway for regulatingangiogenesis. Thus, our observation of an approximately 50%reduction in tube formation by inhibition of the PI3K/Aktpathway in AEG-1 overexpressing HUVECs (Fig. 4) indicatesthat other signaling pathways might partially contribute toAEG-1-induced increased angiogenesis. Future studies are re-quired to clarify the complex mechanisms involved in thisphenomenon.

In conclusion, this study evaluates the direct oncogenic po-tential of AEG-1 as a single gene in nontransformed normalcells. Our data show that AEG-1 plays a crucial role in bothoncogenic transformation and angiogenesis, which are essentialcomponents in tumor cell development, growth, and progressionto metastasis (54). In the contexts of angiogenesis, we defineimportant changes induced by AEG-1 in both HUVECs and

malignant glioma cells, namely upregulation of Ang1 and Tie2,as potential factors underlying pro-angiogenic activity of AEG-1.We further document the importance of PI3K/Akt signaling inmediating pro-angiogenic action of AEG-1. The present studiesprovide mechanistic insights into AEG-1 function and supportthe development of therapeutic strategies that target this gene(or its downstream mediators of transformation and angiogen-esis) by a genetic (antisense or siRNA) or pharmacological(small-molecule) inhibitor approach to develop an effectiverational target-based strategy for the therapy of multiplecancers (54).

Materials and MethodsCell Lines and Culture Conditions. CREF is a specific clone of Fischer F2408 ratembryo fibroblast cells (55). CREF-AEG-1 clone 2, 3, 25, 29, and 30 are hygro-mycin-resistant stable clones obtained after transfection of CREF cells withpcDNA3.1-AEG-1 and selected for growth in 100 �g/mL hygromycin (Invitro-gen). Normal human cells, PHFA, P69 (SV40-immortalized normal humanprostate epithelial cells), and FM516-SV, have been described in ref. 20.Human tumor cells, U87MG and H4 (malignant glioma); DU-145 and PC-3(prostate carcinoma); and HO-1 and C8161 (metastatic melanoma cells), havebeen described (12, 15). Cells resistant to hygromycin were maintained inmedium containing 50 �g/mL hygromycin. HUVEC cells were purchased fromCambrex and maintained according to manufacturer’s instructions.

Xenograft Studies in Athymic Nude Mice. Xenograft studies in nude mice withCREF, CREF-AEG-1 clone 2, 3, 25, 29, and 30 (CREF stably transfected with thehuman AEG-1 cDNA) were performed as described (23, 24). Four weeks afterinjection, mice were killed, tumors were weighed and measured, and tumorswere fixed in 10% neutral buffered formalin for the immunohistochemicalstudies. Statistical significance between the groups was determined by un-paired two-tailed Student’s t test. A P value of � 0.05 was consideredsignificant.

Immunohistochemistry. Formalin-fixed tumors were embedded in paraffin,sectioned, and mounted on glass slides (55). Immunohistochemical stainingwas performed with anti-rabbit AEG-1, Ang1, and MMP-2, and anti-mouseCD31 and HIF1-� antibodies as described in ref. 24.

Recombinant Adenovirus Constructs and siRNAs. The recombinant replication-incompetent Ad.AEG-1 and Ad.DN.Akt and the control and AEG-1 siRNAswere described (13, 20). Tie2 siRNA was purchased from Santa Cruz Biotech-nology. The transfection of siRNA was performed as described in ref. 13.

Invasion Assays. Invasion assays were performed by using 24-well BioCoat cellculture inserts with an 8-�-porosity polyethylene terepthylate membranecoated with Matrigel basement membrane matrix (100 �g/cm2) as describedin ref. 15. The invasive ability of the cells was determined by counting of thecells that had migrated to the lower side of the filter with a microscope(magnification, �100). Experiments were assayed in triplicate, and at least 10fields were counted in each experiment.

Anchorage-Independent Growth Assay in Soft Agar. Anchorage-independentgrowth assays were performed by seeding 1 � 105 cells in 0.4% Noble agar onan 0.8% agar base layer, both of which contained growth medium. Colonieswere counted (�0.1-mm) 2 weeks after seeding, and the data from triplicatedeterminations were expressed as mean � SD.

Western Blot Analysis. Whole-cell lysates were prepared, and Western blotanalysis was done as described in ref. 14. The primary antibodies used wereanti-AEG-1 (chicken polyclonal, 1:2,000), anti-HIF1-� (mouse monoclonal,1:500; AbCam), anti-EF1� (mouse monoclonal, 1:1,000; Upstate Biotechnol-ogy), anti-Tie2, and anti-Ang1 (rabbit polyclonal, 1:500; Santa CruzBiotechnology).

Capillary-Like Tube Formation Assay. The formation of tube-like structures byHUVECs on Matrigel (Chemicon) was performed as described in ref. 56. Theadenovirus-transduced or siRNA-transfected HUVECs (pretreated with cal-cein-AM) were seeded on coated plates at 5 � 104 cells/well in EGM containing2% FBS and incubated at 37 °C for overnight. Cells were observed using aFluorescence microscope (Nikon). Images were captured with a video graphicsystem. The degree of network formation was quantified using Image ana-lyzer (National Institutes of Health Image).

Fig. 6. A hypothetical model of the signal transduction pathways involvedin AEG-1-mediated oncogenic transformation and angiogenesis. PI3K is acti-vated by AEG-1, and AEG-1 induces VEGF and Tie-2/Ang. PI3K is also activatedby growth factors and angiogenesis inducers such as VEGF and angiopoietins.Akt is an essential downstream target of PI3K for mediating angiogenicsignals. Akt activation increases HIF1 expression, which in turn increases VEGFtranscriptional expression. Additionally AEG-1 activates NF-�B, which is in-volved in increased invasion and migration and thus indirectly facilitatestumor angiogenesis. VEGF and VEGFR can form an autocrine loop to regulatetumor angiogenesis.

21304 � www.pnas.org�cgi�doi�10.1073�pnas.0910936106 Emdad et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

27, 2

021

Chorioallantoic Membrane Assay. To detect in vivo tubule formation, weperformed CAM assays as described in ref. 57. H4 glioma cells treated witheither control or AEG-1 siRNA were seeded on the CAM surface of 9-day-oldchick embryos. One week after inoculation, the neovasculature was examinedand photographed.

VEGF-Promoter Assay. H4 cells were seeded at 1 � 105/well in 12-well plates24 h before infection. The adenovirus-transduced or siRNA-transfected cellswere cotransfected with 2 �g pGL3/VEGF promoter vector (55) and 0.5 �gpSV-�-galactosidase (�-gal) vector (Promega), and luciferase and �-gal assayswere performed as described in ref. 14.

Statistical Analysis. All of the experiments were performed at least threetimes. The results are expressed as mean � SD. Statistical comparisons weremade using an unpaired two-tailed Student t test. A P value�0.05 wasconsidered as significant.

ACKNOWLEDGMENTS. The present study was supported in part by NationalInstitutes of Health grant nos. CA134721, CA097318, and NS31492 (to P.B.F.) andawards from the Goldhirsh Foundation for Brain Tumor Research and the DanaFoundation (to D.S.). D.S. is the Harrison Endowed Scholar in the Virginia Com-monwealth University (VCU) Massey Cancer Center. P.B.F. holds the ThelmaNewmeyer Corman Endowed Chair in Cancer Research at the VCU Massey CancerCenter and is a Samuel Waxman Cancer Research Foundation Investigator.

1. Hanahan D, Weinberg RA (2000) The hallmarks of cancer. Cell 100:57–70.2. Hahn WC, Weinberg RA (2002) Modelling the molecular circuitry of cancer. Nat Rev

Cancer 2:331–341.3. Fidler IJ, Ellis LM (1994) The implications of angiogenesis for the biology and therapy

of cancer metastasis. Cell 79:185–188.4. Folkman J (1992) The role of angiogenesis in tumor growth. Semin Cancer Biol

32:65–71.5. Wesseling P, van der Laak JA, Link M, Teepen HL, Ruiter DJ (1998) Quantitative analysis

of microvascular changes in diffuse astrocytic neoplasms with increasing grade ofmalignancy. Hum Pathol 29:352–358.

6. Takahashi Y, Kitadai Y, Bucana CD, Cleary KR, Ellis LM (1995) Expression of vascularendothelial growth factor and its receptor, KDR, correlates with vascularity, metasta-sis, and proliferation of human colon cancer. Cancer Res 55:3964–3968.

7. Weidner N, Semple JP, Welch WR, Folkman J (1991) Tumor angiogenesis and metas-tasis—correlation in invasive breast carcinoma. N Engl J Med 324:1–8.

8. Folkman J (1995) Angiogenesis in cancer, vascular, rheumatoid and other disease. NatMed 1:27–31.

9. Liotta LA, Stetler-Stevenson WG (1991) Tumor invasion and metastasis: An imbalanceof positive and negative regulation. Cancer Res 51:5054s–5059s.

10. Kumar R, Fidler IJ (1998) Angiogenic molecules and cancer metastasis. In Vivo 12:27–34.11. Su ZZ, et al. (2002) Identification and cloning of human astrocyte genes displaying

elevated expression after infection with HIV-1 or exposure to HIV-1 envelope glycop-rotein by rapid subtraction hybridization, RaSH. Oncogene 21:3592–3602.

12. Kang DC, et al. (2005) Cloning and characterization of HIV-1-inducible astrocyteelevated gene-1, AEG-1. Gene 353:8–15.

13. Lee SG, Su ZZ, Emdad L, Sarkar D, Fisher PB (2006) Astrocyte elevated gene-1 (AEG-1)is a target gene of oncogenic Ha-ras requiring phosphatidylinositol 3-kinase and c-Myc.Proc Natl Acad Sci USA 103:17390–17395.

14. Emdad L, et al. (2006) Activation of the nuclear factor kappaB pathway by astrocyteelevated gene-1: Implications for tumor progression and metastasis. Cancer Res66:1509–1516.

15. Sarkar D, et al. (2008) Molecular basis of nuclear factor-kappaB activation by astrocyteelevated gene-1. Cancer Res 68:1478–1484.

16. Emdad L, et al. (2007) Astrocyte elevated gene-1: Recent insights into a novel geneinvolved in tumor progression, metastasis and neurodegeneration. Pharmacol Ther114:155–170.

17. Kikuno N, et al. (2007) Knockdown of astrocyte elevated gene-1 inhibits prostatecancer progression through upregulation of FOXO3a activity. Oncogene 26:7647–7655.

18. Brown DM, Ruoslahti E (2004) Metadherin, a cell surface protein in breast tumors thatmediates lung metastasis. Cancer Cell 5:365–374.

19. Hu G, et al. (2009) MTDH activation by 8q22 genomic gain promotes chemoresistanceand metastasis of poor-prognosis breast cancer. Cancer Cell 15:9–20.

20. Lee SG, et al. (2008) Astrocyte elevated gene-1 activates cell survival pathways throughPI3K-Akt signaling. Oncogene 27:1114–1121.

21. Sutherland HG, Lam YW, Briers S, Lamond AI, Bickmore WA (2004) 3D3/lyric: A noveltransmembrane protein of the endoplasmic reticulum and nuclear envelope, which isalso present in the nucleolus. Exp Cell Res 294:94–105.

22. Britt DE, et al. (2004) Identification of a novel protein, LYRIC, localized to tightjunctions of polarized epithelial cells. Exp Cell Res 300:134–148.

23. Lee SG, et al. (2009) Astrocyte elevated gene-1 contributes to the pathogenesis ofneuroblastoma. Oncogene 28:2476–2484.

24. Yoo BK, et al. (2009) Astrocyte elevated gene-1 regulates hepatocellular carcinomadevelopment and progression. J Clin Invest 119:465–477.

25. Su ZZ, Shi Y, Fisher PB (1997) Subtraction hybridization identifies a progression ele-vated gene PEG-3 with sequence homology to a growth arrest and DNA damageinducible gene. Proc Natl Acad Sci USA 94:9125–9130.

26. Su ZZ, Olsson CA, Zimmer SG, Fisher PB (1992) Transfer of a dominant-acting tumor-inducing oncogene from human prostatic carcinoma cells to cloned rat embryo fibro-blast cells by DNA-transfection. Anticancer Res 12:297–304.

27. Fisher PB, Babiss LE, Weinstein IB, Ginsberg HS (1982) Analysis of type 5 adenovirustransformation with a cloned rat embryo cell line (CREF). Proc Natl Acad Sci USA79:3527–3531.

28. Babiss LE, et al. (1986) Mutations in the E1a gene of type 5 adenovirus alter thetumorigenic properties of transformed cloned rat embryo fibroblast cells. Proc NatlAcad Sci USA 83:2167–2171.

29. Shiojima I, Walsh K (2002) Role of Akt signaling in vascular homeostasis and angio-genesis. Circ Res 90:1243–1250.

30. Vogelstein B, Kinzler KW (2004) Cancer genes and the pathways they control. Nat Med10:789–799.

31. Fisher PB (1984) Enhancement of viral transformation and expression of the trans-formed phenotype by tumor promoters, in Tumor Promotion and Cocarcinogenesis InVitro, Mechanisms of Tumor Promotion, ed Slaga TJ (CRC Press, Boca Raton, FL), pp57–123.

32. Carmeliet P (2000) Mechanisms of angiogenesis and arteriogenesis. Nat Med 6:389–395.

33. Carmeliet P, Jain RK (2000) Angiogenesis in cancer and other diseases. Nature 407:249–257.

34. Papapetropoulos A, et al. (1999) Direct actions of angiopoietin-1 on human endothe-lium: Evidence for network stabilization, cell survival, and interaction with otherangiogenic growth factors. Lab Invest 79:213–223.

35. Koblizek TI, Weiss C, Yancopoulos GD, Deutsch U, Risau W (1998) Angiopoietin-1induces sprouting angiogenesis in vitro. Curr Biol 9:529–532.

36. Stratmann A, Risau W, Plate KH (1998) Cell type-specific expression of angiopoietin-1and angiopoietin-2 suggests a role in glioblastoma angiogenesis. Am J Pathol153:1459–1466.

37. Takahama M, et al. (1999) Enhanced expression of Tie2, its ligand angiopoietin-1,vascular endothelial growth factor, and CD31 in human non-small cell lung carcinomas.Clin Cancer Res 5:2506–2510.

38. Martoglio AM, et al. (2000) Changes in tumorigenesis- and angiogenesis-related genetranscript abundance profiles in ovarian cancer detected by tailored high density cDNAarrays. Mol Med 6:750–765.

39. Shirakawa K, et al. (2001) Absence of endothelial cells, central necrosis, and fibrosis areassociated with aggressive inflammatory breast cancer. Cancer Res 61:445–451.

40. Wang J, et al. (2005) Expressions and clinical significances of angiopoietin-1, -2 and Tie2in human gastric cancer. Biochem Biophys Res Commun 337:386–393.

41. Shim WS, et al. (2002) Angiopoietin 1 promotes tumor angiogenesis and tumor vesselplasticity of human cervical cancer in mice. Exp Cell Res 279:299–309.

42. Machein MR, et al. (2004) Angiopoietin-1 promotes tumor angiogenesis in a rat gliomamodel. Am J Pathol 165:1557–1570.

43. Zadeh G, Koushan K, Pillo L, Shannon P, Guha A (2004) Role of Ang1 and its interactionwith VEGF-A in astrocytomas. J Neuropathol Exp Neurol 63:978–989.

44. Tanaka S, et al. (2002) Tie2 vascular endothelial receptor expression and function inhepatocellular carcinoma. Hepatology 35:861–867.

45. Caine GJ, Blann AD, Stonelake PS, Ryan P, Lip GY (2003) Plasma angiopoietin-1,angiopoietin-2 and Tie-2 in breast and prostate cancer: A comparison with VEGF andFlt-1. Eur J Clin Invest 33:883–890.

46. Yu Y, Varughese J, Brown LF, Mulliken JB, Bischoff J (2001) Increased Tie2 expression,enhanced response to angiopoietin-1, and dysregulated angiopoietin-2 expression inhemangioma-derived endothelial cells. Am J Pathol 159:2271–2280.

47. Zadeh G, et al. (2004) Targeting the Tie2/Tek receptor in astrocytomas. Am J Pathol164:467–476.

48. Lin P, et al. (1997) Inhibition of tumor angiogenesis using a soluble receptor establishesa role for Tie2 in pathologic vascular growth. J Clin Invest 100:2072–2078.

49. Lin P, et al. (1998) Antiangiogenic gene therapy targeting the endothelium-specificreceptor tyrosine kinase Tie2. Proc Natl Acad Sci USA 95:8829–8834.

50. Semenza GL (2003) Targeting HIF-1 for cancer therapy. Nat Rev Cancer 3:721–732.51. Semenza GL (2002) HIF-1 and tumor progression: Pathophysiology and therapeutics.

Trends Mol Med 8:S62–S67.52. Fujikawa K, et al. (1999) Role of PI 3-kinase in angiopoietin-1-mediated migration and

attachment-dependent survival of endothelial cells. Exp Cell Res 253:663–672.53. Papapetropoulos A, et al. (2000) Angiopoietin-1 inhibits endothelial cell apoptosis via

the Akt/survivin pathway. J Biol Chem 275:9102–9105.54. Sarkar D, Emdad L, Lee SG, Yoo BK, Su ZZ, Fisher PB (2009) Astrocyte elevated gene-1

(AEG-1): Far more than just a gene regulated in astrocytes. Cancer Res 69:8529–8535.55. Su ZZ, et al. (1999) PEG-3, a nontransforming cancer progression gene, is a positive

regulator of cancer aggressiveness and angiogenesis. Proc Natl Acad Sci USA 96:15115–15120.

56. Rocnik EF, Liu P, Sato K, Walsh K, Vaziri C (2006) The novel SPARC family memberSMOC-2 potentiates angiogenic growth factor activity. J Biol Chem 281:22855–22864.

57. Pfeifer A, Kessler T, Silletti S, Cheresh DA, Verma IM (2000) Suppression of angiogenesisby lentiviral delivery of PEX, a noncatalytic fragment of matrix metalloproteinase 2.Proc Natl Acad Sci USA 97:12227–12232.

Emdad et al. PNAS � December 15, 2009 � vol. 106 � no. 50 � 21305

MED

ICA

LSC

IEN

CES

Dow

nloa

ded

by g

uest

on

Janu

ary

27, 2

021